GM Volt Forum banner

1 - 20 of 25 Posts

·
Registered
Joined
·
251 Posts
Discussion Starter #1 (Edited)
The emphasis on chemical purity of the materials EEStor plans to use for their capacitor is not the critical parameter needed to characterize the material. The most important parameter is yield. And Defect Free Yield is primarily determined by the killer defect density of the dielectric. Let’s consider a dielectric purity, i.e. defect free, of 99.9998% (EEStor announced a purity of 99.92%). Consider a film that is 0.3 micron thick. A hard defect of 0.2 micron diameter would be a killer defect, i.e. it could not sustain the high local electric field and would be very leaky. Let’s assume that one in ten million impurities is a killer defect. This works out to 0.00143 killer defects per square cm.

Now find the area for a 1F capacitor with a dielectric constant of 2400 and effective thickness of 0.3 micron. If we use a Stapler yield model with this killer defect density and dielectric area, we get a yield of less than 1% for a 1F cap. A defect free yield, such as 1%, is not a manufacturable process. I don’t know the precise details of size, structure, and manufacture of EEStor caps or how purity relates to killer defects. By what this little exercise shows is that getting high yields won’t be easy.
 

·
Senior Member
Joined
·
3,689 Posts
I guess it depends on the amount of individually testable units that make up the ultracap package. If they can quickly test and reject bad units then they should be able to get acceptable final product yields.

At this point it's all conjecture. We have no idea at all what they are doing. Only EEscam knows. We should find out in about 5 years or so. Don't hold your breath.
 

·
Registered
Joined
·
251 Posts
Discussion Starter #4 (Edited)
Darn defects.

Texas,

If they can quickly test and reject bad units then they should be able to get acceptable final product yields.
Lets say your a sneaker manufacturer and you tell your market you will sell a new product at a competitive $50 price based on a purity projected yield of 90%. You find your manufacturing process, even with ultra high purity starting material, has high defects and your actual yield is 1%. The number of people, material costs, and number of plants are 90 times more than what you projected based on material purity. You end with a ginormous pile of expensive scrap. Do you have a manufacturable process and a market with sneakers priced at $450? There are many beautiful ideas spawned in a lab that are unmanufacturable.

One of the reasons the USSR collasped under communism is that even though the were blessed with an abundance of natural resources, their manufacturing operations were so poor that the final products they ended up with were worth less than the sum cost of materials, labor, and energy. A whirl of activity but what came out was not worth as much as the cost of what went in.
 

·
Senior Member
Joined
·
3,689 Posts
Texas,



Lets say your a sneaker manufacturer and you tell your market you will sell a new product at a competitive $50 price based on a purity projected yield of 90%. You find your manufacturing process, even with ultra high purity starting material, has high defects and your actual yield is 1%. The number of people, material costs, and number of plants are 90 times more than what you projected based on material purity. You end with a ginormous pile of expensive scrap. Do you have a manufacturable process and a market with sneakers priced at $450? There are many beautiful ideas spawned in a lab that are unmanufacturable.

One of the reasons the USSR collasped under communism is that even though the were blessed with an abundance of natural resources, their manufacturing operations were so poor that the final products they ended up with were worth less than the sum cost of materials, labor, and energy. A whirl of activity but what came out was not worth as much as the cost of what went in.



Like I said, it's all conjecture. We can have philosophical talks about why the USSR collapsed but that's not going to give us any more information about EEscam's processes. They seem pleased with themselves so we will just have to wait and see. I have a feeling we will be waiting a lot longer than many hope.
 

·
Registered
Joined
·
190 Posts
Let’s consider a dielectric purity, i.e. defect free, of 99.9998% (EEStor announced a purity of 99.92%).
This is from the Press Release:
"The purification of the EEStor, Inc. chemicals has been certified by the same chemical analysis company as EEStor's press release dated January 17, 2007 and now indicates that EEStor has improved its chemical purity to the parts-per-billion range. The aluminum oxide particle coating material purification has been certified to be in the parts-per-trillion level.
"The percent of the constituents crystallized in the CMBT powders ranged from 99.57% to 100.00% with the average being 99.92%.

So you are saying that the dielectric purity is the crystallization percent in the powders? But how easy is it to separate the crystals from the powder? Maybe there is another step that rejects all non-crystallized matter?

Also, I thought the patent said a film of 12.76 microns, are you saying it is only 0.3 microns or are you using 0.3 microns as an example?
 

·
Registered
Joined
·
251 Posts
Discussion Starter #7 (Edited)
Hi Kari K,

So you are saying that the dielectric purity is the crystallization percent in the powders? But how easy is it to separate the crystals from the powder? Maybe there is another step that rejects all non-crystallized matter?
As I indicated, what I see as a critical paramter is dielectric film yield, which is related to defect density. A process can have 100% purity and still have a high electrical defect density. This defect density can affect film integrity and allow damage from subsequent processing. I used impurity numbers to get some handle on magnitude of defect density and then applied a very conservative one in ten million defect to impurity ratio. I suspect the ratio is closer to 1 to 1. I have knowledge of MOS gate dielectrics, their associated defects, and breakdown mechanisms, but I would imagine that the same general principles apply.
Also, I thought the patent said a film of 12.76 microns, are you saying it is only 0.3 microns or are you using 0.3 microns as an example?
A thicker film thickness will give a greater immunity to defects. I went back and reran the numbers (one in ten million defect to impurity ratio) with 12 micron thickness and 5 micron defects. The yield estimate increased from 0.5% to 28%. (a one in a million defect to impur)Thickening the film requires a proportionate increase in area (increased probability of failure) to get the same capacitance. I don't know the nature of the defect size distribution. For a fixed defect size distribution thickening the film will decrease the influence of defects.

The impurity levels listed in the patent are probably goals, and not actual results. I reran this with a parts per trillion Al oxide impurity level and a one in a thousand defect to impurity ratio and got a estimate of 7% yield.

My main point is that the estimate still indicates that high yield won't be easy.
 

·
Registered
Joined
·
256 Posts
I don't fully understand all of this talk but IF the purity of the powder is the go-nogo factor in wether this will work, it leads me to believe that it will eventually work as this issue is attacked and overcome.
 

·
Registered
Joined
·
218 Posts
One way that EEStor is improving purity is to use a "wet" chemical process. See the following patent application:

http://appft1.uspto.gov/netacgi/nph-Parser?Sect1=PTO2&Sect2=HITOFF&u=/netahtml/PTO/search-adv.html&r=1&f=G&l=50&d=PG01&p=1&S1=(("nelson+carl").IN.)&OS=in/("nelson+carl")&RS=IN/("nelson+carl")

Quoted from the patent application:

"Despite the advantages of wet chemical processes, the ceramics industry largely remains reluctant to employ these techniques. Conventional methods for preparing ceramic powders entail mechanical mixing of dry powders of water-insoluble carbonates, oxides, and sometimes silicates, where each constituent of the ceramic composition is carefully selected individually. For example, if the ceramic composition has nine constituents in solid solution, then correspondingly nine starting powders are selected in accordance with the amount of each required for the end product compound. The starting powders are very likely to have different median particle sizes and different particle size distributions. In an attempt to comminute the mixture of powders to a smaller, more uniform particle size and size distribution for each component, the powder mixture is placed in a ball mill and milled for several hours. The milling process generates wear debris from the ball mill itself and, the debris becomes incorporated in the powder mixture. Because of the often wide disparity in particle size among the various commercially available starting powders (and even significant variation in particle size of the same powder from lot to lot), an optimum result from ball milling rarely occurs, and a contamination-free product is never obtained."


"It has been discovered that wet-chemical methods involving the use of water-soluble hydrolytically stable metal-ion chelate precursors and an ammonium oxalate precipitant can be used in a co-precipitation procedure for the preparation of ceramic powders. ...composition-modified barium titanate is one of the ceramic powders that can be produced."
 

·
Registered
Joined
·
158 Posts
The EEStor design is, in reality, an array of smaller ultracapacitors. If one or two elements out of the thousands are defective, it won't matter. You'll get slightly less energy stored, but not enough to make any difference.
 

·
Registered
Joined
·
251 Posts
Discussion Starter #11
If one or two elements out of the thousands are defective, it won't matter. You'll get slightly less energy stored, but not enough to make any difference.
To be equivalent to a battery, a cap has to be able to maintain its charge over at least a few days, preferably a week. This means that the total leakage resistance for a 30F cap has to be greater than about 10 Kohms to maintain charge over a few days. If any individual cell fails and exceeds this, the entire cap would probably be out of warranty from excessive leakage. Also the nature of failure of ceramic caps is not self clearing like foil caps. It is possible that leakage may grow into a hard short.

Let’s say there are a hundred composite cells. Then each cell must have a leakage of greater than 1 Mohm. Reliability is related to defects. We have to allow a margin for the defect to get worse with time. Then a reasonable reliability screening spec for a killer defect would then probably be 10 Mohm per individual cell. This would be very difficult to achieve with a few thousand volts across the cell. The slightest contamination can be be ionized (resistivity of pure water is ~18Mohm cm2/cm) and cause a leakage spec failure.

The yield can be improved substantially with smaller individual (e.g. 100uF) caps. This has to be weighed vs. the higher yield losses from increased packaging, assembly, connectors, and processing from proportionately greater component counts.
 

·
Registered
Joined
·
218 Posts
To be equivalent to a battery, a cap has to be able to maintain its charge over at least a few days, preferably a week. This means that the total leakage resistance for a 30F cap has to be greater than about 10 Kohms to maintain charge over a few days. If any individual cell fails and exceeds this, the entire cap would probably be out of warranty from excessive leakage. Also the nature of failure of ceramic caps is not self clearing like foil caps. It is possible that leakage may grow into a hard short.

Let’s say there are a hundred composite cells. Then each cell must have a leakage of greater than 1 Mohm. Reliability is related to defects. We have to allow a margin for the defect to get worse with time. Then a reasonable reliability screening spec for a killer defect would then probably be 10 Mohm per individual cell. This would be very difficult to achieve with a few thousand volts across the cell. The slightest contamination can be be ionized (resistivity of pure water is ~18Mohm cm2/cm) and cause a leakage spec failure.

The yield can be improved substantially with smaller individual (e.g. 100uF) caps. This has to be weighed vs. the higher yield losses from increased packaging, assembly, connectors, and processing from proportionately greater component counts.
A few tidbits that I've gleaned over the years from EEStor patents:

1) ESU is made of 31,000 smaller caps. This improves yield.

2) I thought that each cap had its own fuse. Ceramic caps definately can fail as a hard short & this would blow the fuse.

3) Self discharge is on the order of 0.1% per month.
 

·
Registered
Joined
·
251 Posts
Discussion Starter #13 (Edited)
"Self discharge is on the order of 0.1% per month" implies a total leakage resistance of ~ 100kohm for 30F. 31,000 individual caps implies a leakage greater than 3,000 Mohms per individual cap under high voltage. The resistivity of aluminum oxide, glass, and barium titanate are 10^14 ohm-cm, 10^12 ohm-cm, and 10^10 ohm-cm, respectively. The EEStor barium titanate is a mixture/doped, so it is lower. So with small (~ mm) individual cap dimensions, the leakage spec seems doable. But this is still challenge for a high volume production item tested at high voltage where ionization of the slightest fingerprint or moisture (pure water is ~18Mohm cm2/cm) would fail leakage spec.
 

·
Registered
Joined
·
218 Posts
"Self discharge is on the order of 0.1% per month" implies a total leakage resistance of 100kohm for 30F. 31,000 individual caps implies a leakage greater than 3 billion ohms per individual cap under high voltage. That's a challenge for a high volume production item. The slightest fingerprint or moisture would fail leakage spec.
I don't find 3GOhms per cap unreasonable.
 

·
Registered
Joined
·
218 Posts
The fact that EEStor's ESU is comprised of standard size ceramic caps is all the more exciting because it means that they could sell them individually for integration into just about anything. Hopefully, they'll be able to reduce the oxide thickness & the voltage rating for portable device applications.
 

·
Senior Member
Joined
·
3,689 Posts
Hummmm. So they don't have one cap to show us? They could put it in a little black box with a couple of meters attached. An oscilloscope showing the charging and discharge cycles. Is that really too much to ask? If they can't show us that, how far away do you really think they are from mass production of a pack that holds 31,000? All working together and tested, ready for shipment to Zenn.

Color me skeptical.
 

·
Registered
Joined
·
218 Posts
Hummmm. So they don't have one cap to show us? They could put it in a little black box with a couple of meters attached. An oscilloscope showing the charging and discharge cycles. Is that really too much to ask? If they can't show us that, how far away do you really think they are from mass production of a pack that holds 31,000? All working together and tested, ready for shipment to Zenn.
Once you can make a single capacitor, putting them together is trivial. You just use solder. World sales of ceramic capacitors in 2008 will be about 1 trillion pieces.
 

·
Senior Member
Joined
·
3,689 Posts
Once you can make a single capacitor, putting them together is trivial. You just use solder. World sales of ceramic capacitors in 2008 will be about 1 trillion pieces.


Well, we can look at two examples that show it's not that easy to take a single component and turn it into a production ready assembly in significant volumes (something EEscam has promised Zenn). How about the Telsa battery back and the Volt's pack? Even though the cells were around for a long time AND in mass production look at how long it took to actually put them in a production package, test and certify (the Volt‘s will take another couple of years). How long would you think that would take any experienced company? Only six months?

Also, are you tell me they would build a volume production line for these caps without having a pilot line or even a laboratory line that was far enough along that they felt confident in buying equipment for a full scale production line? Come on...
 

·
Registered
Joined
·
158 Posts
What EEStor seems to be claiming is that the process to create the module is not that difficult, but making the pieces that go into it is tough. Could be, could not be. But, this is what bothers me, just because making the dielectric powder is stunning difficult, it does not follow that you can wave your hands and make a significant problem with assembly not exist. Maybe they are working the assembly into modules with 'fake' dielectric or something. I sure hope so because it would more than a little traumatic for them (and us) to get the dielectric down, then take another year to package just because they were so focused on the 'hard' stuff.

I also think Tom's persistence in his failure mechanism is misplaced, but I more than agree with him that (and others) the possibility of failure is still high. And, I do share the list's frustration with no samples to demonstrate. Not much choice except to wait though. I hate this.
 

·
Registered
Joined
·
251 Posts
Discussion Starter #20
Risks are exponential and not multiplicative

I also think Tom's persistence in his failure mechanism is misplaced
EEStor uses high power, high energy, high current, high voltage, gigantic field strength, non-linear ultra high permittivity technology, each aspect of which is fraught with stress/risks. They are used in a high vibration, -40 to 125C, environment. Stress acceleration factors generally are not multiplicative, but exponential. Let us list some known failure mechanisms:
Transient high temperature accelerated breakdown from high power density charge/discharge, electromigration of thin film conductors under high electron current density, time-dependent dielectric breakdown from electron trapping at defects within the oxide, molecular bond breaking of dielectrics under high fields, ferroelectric narrow-bandgap leakage; all of these are magnified by defects. They are also exploring the unknown realm of non-linear ultra high permittivity technology. Anybody who has spent time doing development knows that Murphy's and Gell-Mann's laws are real, implacable, and fundamental realities. This is especially true in the realm of high volume production automotive, where sub ppm failure rate specs are required.

Consider this. The actual operating life of most consumer or automotive comonents is not all that great. The cumulative operating life of an automotive idle speed motor (gears are made of plastic) is less than two days. Consider a car engine. On average, most people operate a vehicle less than 1.5 hours a day. These caps, however, will be under high voltage stress continuously, 24 hours/day, 365 days/yr. That's 16X the operating hours of an engine. Consequently, they will require a level of reliability more than a order of magnitude higher than most automotive components. Reliability requirements for automotive are much more severe than military or aerospace. Mission reliability for a jet fighter is only 90%. Would you tolerate that for your car? The reliability requirements for aerospace is, of course, tragic.

My insistence is based on the combination of levels of physical stresses/risks.
 
1 - 20 of 25 Posts
Top